Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A method for processing a 3D scene, comprising: obtaining an input image of the scene captured by a first camera; selecting a reference image of the 3D scene among a set of ambient images of the 3D scene captured from different viewpoints of a second camera under ambient lighting, according to a viewpoint corresponding to said input image, said reference image comprising a reference area representative of a reference diffuse surface captured under said ambient lighting; determining a texture map from said reference image according to an ambient light intensity, the ambient light intensity being determined according to at least one part of said reference area; determining a texture-free map, a value of a pixel of the texture-free image map being a ratio of an intensity at the corresponding pixel in said input image over a value of an intensity at the same pixel in the texture map; and processing said 3D scene using said texture-free map.
This invention relates to 3D scene processing, specifically addressing the challenge of separating texture and lighting information from captured images to improve scene analysis or rendering. The method involves obtaining an input image of a 3D scene captured by a first camera. A reference image is selected from a set of ambient images of the same scene, captured by a second camera from different viewpoints under ambient lighting. The selection is based on matching the viewpoint of the input image. The reference image includes a reference area representing a diffuse surface under ambient lighting conditions. A texture map is generated from the reference image, with its properties determined by the ambient light intensity, which is derived from at least part of the reference area. A texture-free map is then created, where each pixel's value is the ratio of the intensity at the corresponding pixel in the input image to the intensity at the same pixel in the texture map. This texture-free map is used to process the 3D scene, enabling applications such as relighting, texture extraction, or scene reconstruction by isolating surface properties from lighting effects. The approach leverages multiple ambient images to ensure accurate texture and lighting separation, improving the fidelity of subsequent scene processing tasks.
2. The method according to claim 1 , wherein processing said 3D scene comprises determining lighting parameters of the scene.
3. The method according to claim 1 , further comprising scaling said texture-free map with a scaling factor.
4. The method according to claim 3 , wherein said scaling factor is selected such that a number of pixels with saturated intensity value in the scaled texture-free map is below a value.
5. The method according to claim 1 , further comprising registering said input image with respect to the viewpoint of the selected reference image and wherein determining said texture-free map uses said registered input image.
6. An apparatus configured to process a 3D scene, the apparatus comprising a processor and a memory having stored instructions operative, when executed by the processor, to cause the apparatus to perform: obtaining an input image of the scene captured by a first camera, selecting a reference image of the 3D scene among a set of ambient images of the 3D scene captured from different viewpoints of a second camera under ambient lighting, according to a viewpoint corresponding, to said input image, said reference image comprising a reference area representative of a reference diffuse surface captured under said ambient lighting, determining a texture map from said reference image according to an ambient light intensity, the ambient light intensity being determined according to at least one part of said reference area, determining a texture-free map, a value of a pixel of the texture-free image map being a ratio of an intensity at the corresponding pixel in said input image over a value of an intensity at the same pixel in the texture map, and processing said 3D scene using said texture-free map.
This invention relates to processing 3D scenes by removing texture information to enhance surface analysis. The problem addressed is the difficulty in accurately analyzing 3D surfaces due to texture variations caused by ambient lighting conditions. The apparatus includes a processor and memory with instructions to obtain an input image of a 3D scene captured by a first camera. A reference image is selected from a set of ambient images captured by a second camera from different viewpoints under ambient lighting, matching the viewpoint of the input image. The reference image includes a reference area representing a diffuse surface under ambient lighting. A texture map is generated from the reference image based on ambient light intensity, which is determined from the reference area. A texture-free map is then created, where each pixel's value is the ratio of the input image's pixel intensity to the corresponding pixel intensity in the texture map. The 3D scene is processed using this texture-free map, effectively removing texture variations to improve surface analysis accuracy. The system leverages multiple viewpoints and ambient lighting conditions to ensure robust texture removal.
7. The apparatus according to claim 6 , wherein processing said 3D scene comprises determining lighting parameters of the scene.
This invention relates to a system for processing three-dimensional (3D) scenes, particularly for determining lighting parameters within the scene. The system includes a processor configured to analyze the 3D scene data, which may be obtained from various sources such as depth sensors, cameras, or pre-existing 3D models. The processor extracts lighting information by evaluating the scene's geometry, surface properties, and environmental factors to estimate parameters like light source positions, intensities, and colors. This analysis may involve computational techniques such as ray tracing, global illumination modeling, or machine learning-based inference to accurately reconstruct or simulate the lighting conditions present in the scene. The determined lighting parameters can then be used for applications such as virtual reality, augmented reality, computer graphics rendering, or real-world lighting optimization. The system may also incorporate additional processing steps, such as noise reduction or scene segmentation, to enhance the accuracy of the lighting parameter estimation. By automating the extraction of lighting information from 3D scenes, the invention aims to improve the efficiency and realism of digital environments and lighting simulations.
8. The apparatus according to claim 6 , wherein the processor is configured to perform scaling said texture-free map with a scaling factor.
9. The apparatus according to claim 8 , wherein said scaling factor is selected by said processor such that a number of pixels with saturated intensity value in the scaled texture-free map is below a value.
10. The apparatus according to claim 6 , wherein said processor is configured to perform registering said input image with respect to the viewpoint of the selected reference image and wherein determining said texture-free map uses said registered input image.
11. A non-transitory computer-readable storage medium having stored instructions that, when executed by a processor, cause the processor to perform: obtaining an input image of the scene captured by a first camera, selecting a reference image of the 3D scene among a set of ambient images of the 3D scene captured from different viewpoints of a second camera under ambient lighting, according to a viewpoint corresponding to said input image, said reference image comprising a reference area representative of a reference diffuse surface captured under said ambient lighting; determining a texture map from said reference image according to an ambient light intensity, the ambient light intensity being determined according to at least one part of said reference area; determining a texture-free map, a value of a pixel of the texture-free map being a ratio of an intensity at the corresponding pixel in said input image over a value of an intensity at the same pixel in the texture map; and processing said 3D scene using said texture-free map.
12. The non-transitory computer-readable storage medium according to claim 11 , wherein processing the 3D scene comprises determining lighting parameters of the scene.
13. The non-transitory computer-readable storage medium according to claim 11 , further comprising instructions that cause the processor to perform scaling the texture-free map with a scaling factor.
14. The non-transitory computer-readable storage medium according to claim 13 , wherein the scaling factor is selected such that a number of pixels with saturated intensity value in the scaled texture-free map is below a value.
15. The non-transitory computer-readable storage medium according to claim 11 , further comprising instructions that cause the processor to perform registering the input image with respect to the viewpoint of the selected reference image and wherein determining the texture-free map uses the registered input image.
This invention relates to computer vision and image processing, specifically for generating texture-free maps from input images. The problem addressed is the challenge of accurately mapping input images to a reference image while minimizing texture-related distortions, which can interfere with subsequent analysis or visualization. The system involves a non-transitory computer-readable storage medium containing instructions that, when executed by a processor, perform image registration and texture-free map generation. The input image is registered to align with the viewpoint of a selected reference image, ensuring spatial consistency. The texture-free map is then determined using the registered input image, effectively removing or suppressing texture details to focus on structural or geometric features. This process may involve techniques such as edge detection, smoothing, or other texture suppression methods to enhance the clarity of the underlying structure. The invention may also include additional steps such as selecting a reference image from a set of images, where the reference image is chosen based on criteria like image quality or structural clarity. The system may further involve generating a 3D model or performing other downstream tasks using the texture-free map, such as object recognition, surface reconstruction, or medical imaging analysis. The overall goal is to improve the accuracy and reliability of image-based analysis by reducing texture interference while preserving essential structural information.
Unknown
February 23, 2021
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.